Page 1 of 11

Research Methods in Finance

By:

Abdul Qadeer Khan (PhD Scholar & Research Associate)

nd
D-Block 2 Floor, Room # 218
Mohammad Ali Jinnah University, Islamabad
Office: 051-4486701 Ext: 212
Cell: 0333-6487274
dr.aqkhan@live.com
abdul.qadeer@jinnah.edu.pk

Section 1, 2 & 3
Lecture # 3
July 2-4, 2012

In this lecture, we will answer the following questions:

 How to identify and remove the outlier?

 How to diagnose omitted variable?
 Decision about inclusion or removal of variables.

Suppose you are given any data; for example MVR.xls file that has been already circulated.

Steps:

i. Copy data of one variable and remember its observations that are shown in “Status Bar”
with the name of “Count”. In this case observations are 1117, as shown in picture below.

Copyrights2012@Mohammad Ali Jinnah University Islamabad

 Page 2 of 11

ii. Open Eviews, Go to file, New Workfile, select Unstructured and put observations. Then
enter and generate one by one all variables as shown in following equation :

Equation for the Generation of Variable

in Eviews

Variable Name
Syntax for
generate

Genr de=0 Zero

Space
is equal to

Copyrights2012@Mohammad Ali Jinnah University Islamabad

 Page 3 of 11

iii. After the generation of all variables, all variables will be shown in small window as given
below.

iv. Open each variable one by one, click edit and paste the data.
v. Now plot any variable: write equation “plot fta”, following results will be shown:

Copyrights2012@Mohammad Ali Jinnah University Islamabad

 Page 4 of 11

vi. Three abnormal patterns are observed, this is called outlier. Plot the graph of all variables
for the purpose of initial behavior of data. This is most easy way to detect outlier.
vii. Two methods for removing outlier: First method is, keep the cursor on most abnormal
behavior of data. Some digits will be shown; in this case [(171 (4.5)], 171 means row and 4.5
means value. For the purpose of removing these outliers go to excel file and select whole
row and delete it. [Remember do this with the help of economic rationale, that how much
this series can show most negative or most positive behavior]

viii. Second way for identifying and removing outlier is: Open excel file “MVR.xls” select first
variable whole column, go to “sort & filter” icon and choose any smallest to largest or
largest to smallest. In this case I choose smallest to largest. Then click “sort”.

Copyrights2012@Mohammad Ali Jinnah University Islamabad

 Page 5 of 11

ix. You will see all values of D/E (known leverage also) sorted from most negative to most
positive as shown in given below picture. According to economic rationale and with logical
argumentations delete most abnormal values. For deletion, select whole rows and delete it.
Do same processes with all variables till the removal of outliers.

Copyrights2012@Mohammad Ali Jinnah University Islamabad

 Page 6 of 11

x. Another way for identification of outlier is: go to original mvr.xls and populate its values in
Eviews. Go to quick menu and keep cursor on series statistics then click Histogram and Stats.
Small window will be open, write name of variable for example “lev” click OK

xi. Following results will be shown. You can diagnose from descriptive statistics: “Max and Min”
values, which are most abnormal.

Copyrights2012@Mohammad Ali Jinnah University Islamabad

 Page 7 of 11

xii. After removing the outliers from excel files. Kindly generate all variables again in Eviews
with new excel outlier free file.
xiii. Open Eviews and select all variables except “c and resid” [shortcut method for selecting all
variables is, select one variable and press Ctrl+A], Right click go to open as Equation as
shown in picture below. You will see in another small window that all variables are already
written as: dpo g lev p t c , here this is default settings “c” is written in the last but you have
to write after first variable, our first variable is dependant variable and all others are
independent variables.

xiv. After writing the equation as “dpo c g lev p t”, some results will be shown. For the time
being, these results show nothing, we will go to view---Coefficients diagnostics and click on
Redundant Variable Test. Small window will open, write one variable (in picture below we
wrote “t”) and click ok.

Copyrights2012@Mohammad Ali Jinnah University Islamabad

 Page 8 of 11

xv. Now, a small window containing some results will be shown. Kindly check the Probability of
F-Statistic. If F-Statistic value is less than 5% or .05 then it indicates that this variable must
be included in the equation. And shows that variable is “significant and not redundant” and
must be the part of equation. If probability of F-Statistics is greater than 5% or .05 then it
shows that variable is “insignificant and redundant” and must not be the part of the
equation. In simple words, Redundancy Test helps us that which variable must be/must not
be included.

Copyrights2012@Mohammad Ali Jinnah University Islamabad

 Page 9 of 11

xvi. [Remember: Redundancy Test and Omitted Variable Test shows the same concepts; for
example: We regress the equation with all variables to find out which variable must or must
not be the part of equation, then we decide to check the probability of F-Statistics, If
probability of F-Statistics is less than 5% or .05 then redundancy test recommend that this
variable must be the part of the equation otherwise not. Now, omitted variable Test: we
regress all variable except dropping deliberately one independent variable from the
equation. We do the process of Omitted Variable Test. Process is same as redundancy test,
difference is: only omit one variable, suppose: dpo c g lev p (t), we omit “t” and equation
will be : dpo c g lev p, Some results will be shown as:

Copyrights2012@Mohammad Ali Jinnah University Islamabad

 Page 10 of 11

xvii. Now we check the probability of “c” which is basically constant, if the P-Value of c is less
than 5% or .05 then this is the indication that there is any other variable which has been
omitted or missed. Effect of that omitted variable comes in ‘c’ which is constant in nature.
So, we will go to view---coefficients diagnostics click omitted variable test then write the
name of omitted variable in small window. Results will be displayed on your Eviews as:

Copyrights2012@Mohammad Ali Jinnah University Islamabad

 Page 11 of 11

We will check the probability of F-Statistics, if the probability is less than 5% or .05 then this is the
indication that this variable is omitted and has significantly impact and must be included. Suppose if we
include all the missing variables but again ‘c’ comes significant then this shows that we have to revise
economic rationale, maybe there is any other aspect or facet which is not being captured till now. After
revising economic rationale and economic locality we will regress again the equation.

Our basic objective was: how to detect outlier, its removal, check the redundancy test and omitted
variable test.

Note: Comments for improvement will be highly appreciated. Thanks.

Abdul Qadeer Khan

Copyrights2012@Mohammad Ali Jinnah University Islamabad